GoEmotions Dataset: Dataset Re-Classification

goemotions-dataset-dataset-re-classification.ipynb

{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/gist/owahltinez/9c765f3596490cba622313e8276b5f7e/goemotions-dataset-dataset-re-classification.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "c-6AaDi8w6n9"
      },
      "outputs": [],
      "source": [
        "#@title Install Dependencies\n",
        "!pip install -q \"tensorflow==2.9.*\"\n",
        "!pip install -q \"tensorflow-text==2.9.*\""
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {
        "id": "_tOll1k9jIPV"
      },
      "outputs": [],
      "source": [
        "#@title Downloading and re-classifying the sentiment140 dataset\n",
        "import csv\n",
        "import pathlib\n",
        "import urllib.request\n",
        "import tensorflow as tf\n",
        "import tensorflow_datasets as tfds\n",
        "import tensorflow_text as tftxt\n",
        "from tqdm.notebook import tqdm\n",
        "\n",
        "emotions = urllib.request.urlopen(\n",
        "    'https://raw.githubusercontent.com/google-research/google-research'\n",
        "    '/master/goemotions/data/emotions.txt').read().decode('utf8').split('\\n')\n",
        "\n",
        "batch_size = 128\n",
        "ds_orig = tfds.load('sentiment140', split='train')\n",
        "output_path = pathlib.Path('.') / 'sentiment140_goemotions.csv'\n",
        "\n",
        "# Download and load our pretrained classifier.\n",
        "url_root = 'https://huggingface.co/owahltinez/goemotions_bert/resolve/main'\n",
        "url_model = f'{url_root}/bert_model.tar.gz'\n",
        "tar_path = tf.keras.utils.get_file(origin=url_model, extract=True)\n",
        "model_path = pathlib.Path(tar_path) / '..' / 'bert_model'\n",
        "classifier = tf.keras.models.load_model(model_path)\n",
        "\n",
        "# Reclassify the dataset by mapping the original text to fine-grained emotions.\n",
        "ds_emo = ds_orig.batch(batch_size).map(lambda x: x['text'])\n",
        "ds_emo = ds_emo.map(lambda x: (x, tf.math.argmax(classifier(x), axis=1)))\n",
        "batch_count = ds_emo.cardinality().numpy()\n",
        "\n",
        "# Save the resulting dataset into a CSV file.\n",
        "with open(output_path, 'w') as f:\n",
        "  writer = csv.writer(f)\n",
        "  writer.writerow(['sentence', 'label'])\n",
        "  for x_batch, y_batch in tqdm(iter(ds_emo), total=batch_count):\n",
        "    for x, y in zip(x_batch, y_batch):\n",
        "      label = emotions[y.numpy()]\n",
        "      sentence = x.numpy().decode('utf8')\n",
        "      writer.writerow([sentence, label])"
      ]
    }
  ],
  "metadata": {
    "accelerator": "GPU",
    "colab": {
      "collapsed_sections": [],
      "provenance": [],
      "authorship_tag": "ABX9TyMX3j4ACyvBieE+uUKamzc+",
      "include_colab_link": true
    },
    "gpuClass": "standard",
    "kernelspec": {
      "display_name": "Python 3",
      "name": "python3"
    },
    "language_info": {
      "name": "python"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 0
}